fix(ci): allowlist estimating tier vocab, utf-8 test reads, version bump + review fixes

Author: elijahr

.version

@@ -1 +1 @@
-0.76.0
+0.76.1

CHANGELOG.md

@@ -9,6 +9,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- **Project-standards discovery and enforcement in develop.** Add generic
+  project-standards (governance-doc) discovery and enforcement to the develop
+  skill.
+
 - **Embarrassment sweep — codified pre-PR diff-hygiene gate.** The develop
   4.6.5 pre-PR gate (`commands/feature-implement.md`) previously validated only
   that claims were true; it now also runs the named "embarrassment sweep" over

docs/commands/feature-discover.md

@@ -2,8 +2,6 @@
 
 ## Workflow Diagram
 
-Now generating the diagrams. Reading is complete — producing Phase 3 (generation) and Phase 4 (verification) output now.
-
 ---
 
 ## Diagram 1: Phase 1.5 Overview

docs/diagrams/commands/feature-discover.md

@@ -1,8 +1,6 @@
 <!-- diagram-meta: {"source": "commands/feature-discover.md", "source_hash": "sha256:e574ba39a04ec1119b5cc30fdd3110bbb243171654d01962f8d43bf0707f2998", "generated_at": "2026-06-11T01:48:52Z", "generator": "generate_diagrams.py"} -->
 # Diagram: feature-discover
 
-Now generating the diagrams. Reading is complete — producing Phase 3 (generation) and Phase 4 (verification) output now.
-
 ---
 
 ## Diagram 1: Phase 1.5 Overview

scripts/check_removed_mode_tokens.py

@@ -87,6 +87,11 @@ class AllowEntry:
         anchor=None,
         reason="archived sub-orchestrator body (historical), path-allowlisted",
     ),
+    AllowEntry(
+        path_glob="skills/estimating-tickets/fleet-effort.md",
+        anchor="NET-NEW COMPLEX",
+        reason="estimating-tickets complexity speed-class vocabulary, not a develop tier-classifier",
+    ),
 )
 
 # --- Allowlist (b): removed-mode vocabulary that legitimately survives --------

tests/test_develop_gates.py

@@ -22,43 +22,38 @@
 
 _ROOT = Path(__file__).parent.parent
 
-# Source fixtures are read lazily (not at import/collection time) so pytest
-# collection performs no file I/O. Each reader is cached, so repeated access
-# within a run reads the file at most once. Bare-name module-level constants
-# would force collection-time reads — and a module-level __getattr__ would NOT
-# help, because __getattr__ only fires for MISSING globals accessed via
-# attribute lookup, never for bare-name references inside this same module
-# (those raise NameError directly). Hence: explicit cached reader functions.
+# Source fixtures are read lazily via cached reader functions so pytest
+# collection performs no file I/O; each file is read at most once per run.
 
 
 @lru_cache(maxsize=None)
 def FEATURE_RESEARCH() -> str:
-    return (_ROOT / "commands" / "feature-research.md").read_text()
+    return (_ROOT / "commands" / "feature-research.md").read_text(encoding="utf-8")
 
 
 @lru_cache(maxsize=None)
 def FEATURE_DISCOVER() -> str:
-    return (_ROOT / "commands" / "feature-discover.md").read_text()
+    return (_ROOT / "commands" / "feature-discover.md").read_text(encoding="utf-8")
 
 
 @lru_cache(maxsize=None)
 def FEATURE_DESIGN() -> str:
-    return (_ROOT / "commands" / "feature-design.md").read_text()
+    return (_ROOT / "commands" / "feature-design.md").read_text(encoding="utf-8")
 
 
 @lru_cache(maxsize=None)
 def FEATURE_IMPLEMENT() -> str:
-    return (_ROOT / "commands" / "feature-implement.md").read_text()
+    return (_ROOT / "commands" / "feature-implement.md").read_text(encoding="utf-8")
 
 
 @lru_cache(maxsize=None)
 def DEVELOP_SKILL() -> str:
-    return (_ROOT / "skills" / "develop" / "SKILL.md").read_text()
+    return (_ROOT / "skills" / "develop" / "SKILL.md").read_text(encoding="utf-8")
 
 
 @lru_cache(maxsize=None)
 def CODE_REVIEW_GIVE() -> str:
-    return (_ROOT / "commands" / "code-review-give.md").read_text()
+    return (_ROOT / "commands" / "code-review-give.md").read_text(encoding="utf-8")
 
 
 def _flags(research=False, design=False, infrastructure=False):